Metabolic Flux Analysis for Energy Production in Physiological Condition

Synopsis

This software is used for 13C metabolic flux analysis based on a multi-tissue model to analyze nutrition utilization in physiological condition. The 13C tracing data in data_collection.xlsx (low infusion data from Hui et al, 2017) and data_collection_Dan.xlsx (high infusion data) are used as input. Users can choose model to run and corresponding results will be written to new_models folder.

Data

All ¹³C-isotope labeling data are in data/data_collection.xlsx(low-infusion data) and data/data_collection_from_Dan.xlsx(high-infusion data). Concentration data of glucose, lactate and insulin during infusion are in data/related_data_from_dan.xlsx.

Requirements

This software is developed and tested on Python 3.6. It also relies on following Python packages:

Packages	Version has been tested
numpy	1.16
scipy	1.3
matplotlib	2.2
tqdm	4.30
python-ternary	1.0
xlrd	1.1

For convenience, an out-of-the-box Docker image is provided to run this code. This docker image is tested on docker-ce in Ubuntu with Docker version 19.03.1. (See Usages part for details.)

Usages

Docker (recommended)

First install an appropriate Docker version (see here for details). Then a model could be executed by the following script:

MODEL=model1_m5
TARGET_PATH=/your/path/to/output
cd $TARGET_PATH
docker run -it --rm --name python_$MODEL -v `pwd`:/Lactate_MFA/new_models \
  locasalelab/lactate_mfa:latest $MODEL --test_mode

In this script, you could modify the value of MODEL to the name of your target model, and modify the value of TARGET_PATH to the path that you want to output results. Available model name is listed in following section. BE CAREFUL that the target path would be visited as root account in container! The flag --test_mode or -t makes the code run quickly in a test mode, and could be removed to run a formal mode. The formal running takes tens of hours.

System Python interpreter

This script could also be executed as a raw Python project. Make sure Python 3.6 and all required packages are correctly installed. First switch to a target directory and download the source code:

git clone https://github.com/LocasaleLab/Lactate_MFA

Switch to the source direct, add PYTHONPATH environment and run the new_model_main.py:

MODEL=model1_m5
cd Lactate_MFA
export PYTHONPATH=$PYTHONPATH:`pwd`
python src/new_model_main.py $MODEL --test_mode

Similar with Docker, you could modify the value of MODEL to the name of your target model. Final results will be written to the new_models folder in current directory. Available model name is listed in following section. The flag --test_mode or -t could also be removed to run a formal mode.

Parameters

-p, --parallel_num:

Number of parallel processes. If not provided, it will be selected according to CPU cores.

-t, --test_mode:

Whether the code is executed in test mode, which means less sample number and shorter time (several minites).

List of models

Model name in this script	Model name in methods	Source tissue	Sink tissue	Circulating metabolites	Data source	Description
model1	Model A	Liver	Heart	Glucose; Lactate	Low-infusion glucose data: mouse M1	Basic two-tissue model.
model1_unfitted	Model A	Liver	Heart	Glucose; Lactate	Low-infusion glucose data: mouse M1	Unfitted result of basic two-tissue model, as the negative result of fitting.
model1_all	Model A	Liver	All 8 tissues	Glucose; Lactate	Low-infusion glucose data: mouse M1	Basic two-tissue model with different sink tissues.
model1_all_m5	Model A	Liver	All 8 tissues	Glucose; Lactate	Low-infusion glucose data: mouse M5	Basic two-tissue model with different sink tissues and different mouse data.
model1_all_m9	Model A	Liver	All 8 tissues	Glucose; Lactate	Low-infusion glucose data: mouse M9	Basic two-tissue model with different sink tissues and different mouse data.
model1_all_lactate	Model A	Liver	All 8 tissues	Glucose; Lactate	Low-infusion lactate data: mouse M3	Basic two-tissue model with different sink tissues and different infusion data.
model1_all_lactate_m4	Model A	Liver	All 8 tissues	Glucose; Lactate	Low-infusion lactate data: mouse M4	Basic two-tissue model with different sink tissues and different infusion data.
model1_all_lactate_m10	Model A	Liver	All 8 tissues	Glucose; Lactate	Low-infusion lactate data: mouse M10	Basic two-tissue model with different sink tissues and different infusion data.
model1_all_lactate_m11	Model A	Liver	All 8 tissues	Glucose; Lactate	Low-infusion lactate data: mouse M11	Basic two-tissue model with different sink tissues and different infusion data.
parameter	Model A	Liver	Heart	Glucose; Lactate	Low-infusion data: mouse M1	Sensitivity analysis of data and other constraint fluxes.
model6	Model B	Liver	Skeletal muscle	Glucose; Lactate	High-infusion data: mouse M1	Two-tissue model with high-infusion data in different mouse strain.
model6_unfitted	Model B	Liver	Skeletal muscle	Glucose; Lactate	High-infusion data: mouse M1	Unfitted result of two-tissue model with high-infusion flux, as the negative result of fitting.
model6_m2	Model B	Liver	Skeletal muscle	Glucose; Lactate	High-infusion data: mouse M2	Two-tissue model with high-infusion data in different mouse strain.
model6_m3	Model B	Liver	Skeletal muscle	Glucose; Lactate	High-infusion data: mouse M3	Two-tissue model with high-infusion data in different mouse strain.
model6_m4	Model B	Liver	Skeletal muscle	Glucose; Lactate	High-infusion data: mouse M4	Two-tissue model with high-infusion data in different mouse strain.
model3	Model D	Liver	Heart	Glucose; Pyruvate; Lactate	Low-infusion glucose data: mouse M1	Two-tissue model with three circulatory metabolites.
model3_unfitted	Model D	Liver	Heart	Glucose; Pyruvate; Lactate	Low-infusion glucose data: mouse M1	Unfitted result of two-tissue model with three circulatory metabolites, as the negative result of fitting.
model3_all	Model D	Liver	All 8 tissues	Glucose; Pyruvate; Lactate	Low-infusion glucose data: mouse M1	Two-tissue model with three circulatory metabolites and different sink tissues.
model5	Model C	Liver	Heart; Skeletal muscle	Glucose; Lactate	Low-infusion data: mouse M1	Three-tissue model.
model5_comb2	Model C	Liver	Brain; Skeletal muscle	Glucose; Lactate	Low-infusion data: mouse M1	Three-tissue model.
model5_comb3	Model C	Liver	Heart; Brain	Glucose; Lactate	Low-infusion data: mouse M1	Three-tissue model.
model5_unfitted	Model C	Liver	Heart; Skeletal muscle	Glucose; Lactate	Low-infusion data: mouse M1	Unfitted result of three-tissue model, as the negative result of fitting.
model7	Model E	Liver	Skeletal muscle	Glucose; Pyruvate; Lactate	High-infusion data: mouse M1	Two-tissue model with three circulatory metabolites and high-infusion data.
model7_unfitted	Model E	Liver	Skeletal muscle	Glucose; Pyruvate; Lactate	High-infusion data: mouse M1	Unfitted result of two-tissue model with three circulatory metabolites and high-infusion flux, as the negative result of fitting.

Result display

Results generated in computation are plotted to figures by functions in Jupyter Notebook file data_process.ipynb. Figures in paper can also be found in this file.

Contributors

Shiyu Liu

• http://github.com/liushiyu1994

License

This software is released under the MIT License.